In a fuel injection system of this type, the piezo injector has a piezoelectric actuator and a nozzle needle which can be moved by the piezoelectric actuator. The piezoelectric actuator is actuated by means of an electrical actuation signal. On the basis of this actuation signal, the piezoelectric actuator is lengthened in such a way that a mechanical reciprocal movement is brought about. This reciprocal movement moves the nozzle needle by means of which injection holes in a nozzle unit are opened to a greater or lesser extent, in order to be able to inject a desired quantity of fuel, dependent on the specified electrical actuation signal, through the piezo injector into a cylinder of the respective motor vehicle. When the nozzle needle moves, it can arrive at its mechanically predefined needle stroke limitation, i.e. at its needle stop position.
If the electrical actuation of the piezo injector takes place in such a way that the charging time lasts up to the start of the discharging, it is then possible, given a predefined pressure, for various opening times in the chronological vicinity of the needle stop to provide the same quantity of fuel, with the result that there is an ambiguity present in the injection quantity/injection period characteristic diagram. In this context, the injected quantity of fuel can also drop given an increase in opening time. The reason for this is bouncing of the nozzle needle at the stop and the associated different closing speeds dependent on the actuation signal. The term opening time is understood below always to mean the time period which starts with the activation of the injector drive, i.e. the charging of the piezo actuator for the purpose of opening the injection holes, and ends with the start of deactivation of the injector drive, i.e. the discharging of the piezo actuator for the purpose of closing the injection holes.
The specified ambiguity in the injection quantity/injection period characteristic diagram is illustrated in FIG. 1. This shows a diagram in which the injection period TI is plotted along the abscissa and the injection quantity Q is plotted along the ordinate. From this diagram it is apparent that when a pressure value p=1200 bar is present a requested injection quantity of 16 mm3 with three different injection periods t1, t2 and t3 can be implemented, wherein in the exemplary embodiment shown t1=0.34 ms, t2=0.36 ms and t3=0.42 ms.
In the case of TI=t1=0.34 ms, the piezo actuator is energized for 0.34 ms, with the result that the nozzle needle does not yet bounce or bounces only very little at the stop and then closes.
In the case of TI=t2=0.36 ms, the piezo actuator is energized for 0.35 ms, which gives rise to strong bouncing. This bouncing accelerates the closing process to such an extent that despite a relatively long opening time no additional quantity of fuel is injected.
In the case of TI=t3=0.42 ms, the injection period is just long enough to equalize the bouncing back of the needle through an extended opening time.
In this example in which the pressure p is 1200 bar, the piezo actuator is lengthened in the time interval between TI=0.4 ms and TI=0.6 ms to such an extent that the piezo injector opens again completely from TI=0.6 ms. This is the cause of the high gradient starting from the minimum. From approximately TI=0.6 ms, the piezo injector is finally completely throttled, as a result of which the injection quantity only then depends on the injection period.
Owing to the fact that there is no clear relationship between the injection quantity Q and the injection period TI, a plurality of injection periods can be assigned to one requested quantity of fuel. Consequently, it is not possible to regulate the injection quantity with the injection period as a manipulated variable. A regulator would not regulate in a robust fashion in the surroundings of the needle stop but rather tend to swing.
The specified ambiguity in the injection quantity/injection period characteristic diagram is caused by the impetus of the nozzle needle which causes the nozzle needle to recoil somewhat as a result of striking against the needle stroke limitation or the needle stop position. This impetus is determined by the energy applied to the nozzle needle, which energy depends directly on the current and charging time of the needle drive given a predefined actuation power. The charging time describes the time during which the drive, i.e. the piezo actuator, is energized.
For the so-called full stroke operating mode, the nominal charging time tnom,charge of the drive is dimensioned to open the piezo injector at a given current in such a way that the needle reliably reaches its stop position. Said needle is at maximum several 100 μs and is independent of relatively long opening times. If relatively short opening times are desired, the charging time is set to be equal to the opening time. The form of the charging current is independent of the charging time. The charging current is cut off at the end of the charging time.
In the case of the known method described above for actuating a piezo injector the following relationships apply:tcharge(p)=tnom,charge(p) for TI≧tnom,charge  (equation 1)tcharge(p)=TI for TI<tnom,charge  (equation 2).
It is already known to counteract bouncing and the ambiguity of the injection quantity/injection period characteristic diagram caused thereby through long actuation times given a low actuation power.
As a result, the speed of the nozzle needle is reduced, which in turn reduces the bouncing of the nozzle needle at the needle stroke limitation. Furthermore, part of the energy is applied only after the needle stroke limitation has been reached. In such a procedure, the bouncing can be reduced sufficiently only for very low actuation power levels and long actuation times associated therewith.
Further disadvantages of such a slow energy supply, i.e. a long charging time of the piezo actuator, are, in particular, the slow passage through the seat throttling and the late acquisition of the energy which is necessary for completely dethrottling or reaching the needle stop. Furthermore, in the case of such a reduction in the bouncing the strength of an important signal, namely the needle stop signal, is reduced. This makes detection of the striking of the needle against its stop position more difficult. A further disadvantage of such a slow supply of energy is the fact that the needle remaining for a long time in the throttling makes reaching the requested minimum quantity accuracy more difficult.